2026 Fall Conference Poster Competition Selected Abstracts

Both thymocyte and T-fib-derived iPSC lines exhibited characteristic pluripotent stem cell morphology, expressed pluripotency markers, maintained normal karyotypes, and demonstrated trilineage differentiation potential. Importantly, thymic-derived iPSCs successfully generated downstream cellular therapy candidates, including endothelial cells, cardiomyocytes, and thymic epithelial progenitors, highlighting their versatility for regenerative medicine applications. Methylation profiling via microarray revealed that early-passage thymic-iPSCs retained DNA methylation signatures reflective of their cell of origin, consistent with epigenetic memory, whereas these differences diminished with continued passaging.

These findings establish discarded neonatal thymus tissue as a valuable source of patient-specific iPSCs for regenerative medicine. In the future, thymic-derived iPSCs could enable autologous cellular therapies for pediatric patients undergoing congenital heart surgery, including generation of replacement thymic tissue to restore immune function and cardiac cell therapies to repair congenital heart defects.

Methods. Nine single-cell RNA-sequencing libraries spanned two experiments. A single PBMC donor was co-cultured <24 hours with either autologous or HLA-mismatched iPSC-ECs, each arm with its own EC-alone control, permitting a difference-of-differences (ΔΔ) contrast that cancels cell line baseline. An independent donor pair tested replication. Because each condition is a single library, we report pseudobulk effect sizes, split-fold sign stability, and cross-pair directional concordance. Pathways were scored against MSigDB, PROGENy, and CollecTRI.

Results. Allogeneic co-culture induced IFN-γ associated pathways, chemokines (CXCL9/10/11), adhesion molecules (ICAM1, VCAM1), and the upstream STAT1-IRF1 signature. Autologous co-culture induced similar changes with reduced magnitude. Of 207 genes with >2-fold all-specific induction, MHC class II separated the arms most sharply (CD74 ΔΔ +3.2, HLA-DRA ΔΔ +2.6). Every immune lineage showed matching interferon signatures. The allogeneic endothelial response replicated across donor pairs: 95% of 1690 genes agreed in direction.

Conclusions. A purified iPSC-EC/PBMC platform recapitulated the interferon axis of early allorecognition with high resolution and independent replication. Because the endothelium is common to vascularized allografts, signatures defined in this system offer a mechanistic reference against which existing rejection signatures can be interpreted. With one library per condition and one autologous arm, effect size and cross-pair concordance establish the analytic baseline for the “Rejectome”.

1Department of Cell and Regenerative Biology, School of Medicine and Public Health,
University of Wisconsin – Madison, Madison, WI, 53705, USA.

Regenerative factors stimulate diverse cardiac cell types to cooperate in exquisite harmony to ensure heart regeneration. We previously identified Interleukin 11 (Il11) as a potent regenerative factor in zebrafish hearts, capable of stimulating cardiomyocyte (CM) proliferation and promoting vasculogenesis in the absence of injury. To dissect how heterogeneous cardiac cell types respond to regenerative stimuli, we performed and analyzed single cell RNA sequencing (scRNA-seq) following il11a overexpression (il11aOE). Our scRNA-seq analysis revealed a distinct subset of cardiomyocytes (CMs) that emerges upon il11aOE. This il11-responsive CM cluster is highly enriched for cell cycle-related gene expression, supporting that il11a acts as a mitogen for CMs. We are currently investigating transcription factors that mediate il11a-triggered CM proliferation. il11aOE also increases the number of ACTA2⁺ primordial CMs by approximately two-fold. Our scRNA-seq analysis further showed that crb2a is highly enriched in ACTA2+ primordial CMs. Notably, Crb2 polarity in these primordial CMs is significantly shifted from apical to junctional localization upon il11a induction, suggesting CM polarity is remodeled in response to il11a. In addition to CMs, il11aOE markedly expands epicardial (Epi) and epicardial-derived cells (EPDCs), giving rise to 6 distinct sub-clusters. These heterogeneous populations include Periostin⁺ (Postn) fibroblasts and ptx3a⁺ Epi/EPDC subsets. The ptx3a+ Epi/EPDC exhibits high expression of angiogenic factors, suggesting a role in activating coronary vessels and promoting vasculogenesis. Overall, these results provide molecular insight into how regenerative cytokine signaling reshapes the transcriptional states and cellular composition of the heart, revealing coordinated cell-type–specific programs that enable cardiac tissues to mount a regenerative response.

AIM: We investigate iPSC-derived epicardial cells in hypoxic conditions to mimic MI. We aim to identify a therapeutic target which could aid EMT peri-infarct. One of our top candidates is NOTCH3, which plays an important role in cardiac development.

METHODOLOGY: We mimic hypoxia to model MI in vitro via 1. Culturing cells in a hypoxia chamber and 2. With cobalt chloride, a common hypoxia inducing agent. We then compare expression of EMT markers, hypoxia induction pathway markers, and epicardial maintenance identity markers via ICC and flow cytometry. We look at both changes after hypoxia culture and after simulated reperfusion injury by re-exposing cells to ambient oxygen conditions. We then investigate changes in these markers in the presence of NOTCH3.

RESULTS: We found differences in EMT expression in the presence of NOTCH3 and hypoxia, including decreased epicardial marker expression. We also found differences in hypoxia marker expression.

CONCLUSION/DISCUSSION: Our results demonstrate a potential for the NOTCH3 pathway to be targeted peri-infarct as a means of promoting cardiac regeneration. Furthermore, as there is no published literature investigating the effects of hypoxia in human epicardial/ epicardial-like cells in vitro, we have also established a new platform for studying EMT.

Method. We used induced pluripotent stem cells (iPSCs) from 9p21.3 risk and non-risk donors to create isogenic lines with a full haplotype deletion, generating four genotypes: risk WT, risk KO, non-risk WT, and non-risk KO. Using two independent iPSC lines per genotype (8 lines total), we conducted a differentiation to VSMC and employed single-cell transcriptomic profiling. We then performed co-embedding and comparison with publicly available human coronary artery scRNAseq datasets, validated the hits from our analysis by immunocytochemistry, and performed functional assays to test the impact of the transcriptional changes. Finally, we used knockdown and overexpression experiments to further examine the impact of key genes identified in our analysis.

Results. Our study shows that iPSC-VSMCs carrying the 9p21.3 risk haplotype adopt an osteochondrogenic state. The altered transcriptional profile is characterized by decreased migration capacity and increased calcification, revealing a phenotype closely resembling fibrochondrocytes, which are an osteochondrogenic state known to contribute to CAD pathogenesis. Additionally, we observe a strong transcriptional overlap between iPSC-VSMCs and native human VSMCs from heart arteries, validating the utility of our model to study human disease. Finally, we identify two novel 9p21.3 risk-biased signature genes, LIMCH1 and CRABP1, which appear to be critical in defining the risk state.

Conclusions. Our study demonstrates that 9p21.3 drives disease-relevant phenotypic modulation in VSMCs consistent with an osteochondrogenic-like state called fibrochondrocytes. We observe that the 9p21.3 risk region exerts its influence in a cell type-specific and cell-autonomous manner, promoting CAD-relevant phenotypes like arterial stiffening and calcification. This work contributes to our knowledge of how 9p21.3 can increase risk of CAD, thus aiding efforts to mitigate genetic risk for the disease.

Background: Arrhythmogenic cardiomyopathy (ACM) is characterized by ventricular arrhythmias that can precede detectable structural remodeling, yet mechanisms underlying this concealed electrical phase remain poorly understood. Most human induced pluripotent stem cell (hiPSC) models focus on cardiomyocyte (CM)-intrinsic defects, leaving the contribution of cardiac fibroblasts (CFs) unresolved. We studied a patient carrying PKP2 p.L709F who presented with exercise-associated sustained ventricular tachycardia despite normal cardiac MRI and later developed right ventricular remodeling.

Hypothesis: PKP2 p.L709F produces a CM-intrinsic conduction defect that is further modulated by pathogenic interactions with second heart field-derived cardiac fibroblasts (SHF-CFs).

Methods: Patient-derived PKP2 p.L709F hiPSCs and a CRISPR/Cas9-corrected isogenic control were differentiated into hiPSC-CMs and SHF-CFs. CMs were studied in monoculture and in micropatterned co-cultures containing mutant or gene-corrected SHF-CFs. An independent control hiPSC-CM line was also co-cultured with mutant or corrected SHF-CFs. Electrical propagation was quantified by FluoVolt optical mapping, and junctional connexin-43 (Cx43) organization was assessed by immunocytochemistry.

Results: Mutant hiPSC-CMs exhibited slower conduction than gene-corrected hiPSC-CMs in monoculture, demonstrating a CM-intrinsic phenotype. This defect was further exacerbated in line-matched mutant co-cultures. Genotype-mismatched experiments revealed an independent fibroblast contribution: mutant SHF-CFs impaired electrical propagation, whereas gene-corrected SHF-CFs partially rescued conduction in mutant hiPSC-CMs. Mutant SHF-CFs also reduced conduction in an independent control hiPSC-CM line. Mutant line-matched co-cultures exhibited reduced junctional Cx43 organization compared with corrected controls. Mutant SHF-CFs were also associated with reduced upstroke velocity, consistent with impaired intercellular coupling and/or CM excitability.

Conclusions: These findings identify both CM-intrinsic and fibroblast-dependent mechanisms of electrical dysfunction in PKP2-associated ACM and reveal a prominent contribution of SHF-CF genotype to conduction. This human multicellular model enables mechanistic investigation of concealed-phase arrhythmogenesis and provides a platform for identifying early therapeutic targets before overt structural remodeling.

The LMNA gene encodes nuclear lamins A and C, intermediate filaments with roles in gene regulation, chromatin organization, and nuclear stability. LMNA mutations lead to a broad set of inherited diseases called laminopathies. Among these, cardiac laminopathies present early with conduction defects and arrhythmias, and progressive cardiomyopathy. The major cellular phenotype has focused on characterization of cardiomyocytes, yet cardiac fibroblasts may also contribute to disease progression. The specific role of cellular crosstalk between cardiac fibroblasts and cardiomyocytes remains unclear.

We hypothesize that the arrhythmic phenotype observed in LMNA mutations is dependent on cellular crosstalk between hiPSC-cardiomyocytes (CMs) and hiPSC-cardiac fibroblasts (CFs) and that co-culture groups will exhibit differentially expressed intercellular signaling mediators that correlate with electrophysiologic and mechanical dysfunction. To test this hypothesis, we introduced heterozygous LMNA+/E203K mutation, a LMNA mutation associated with cardiac laminopathy, using CRISPR-gene editing into a non-diseased control iPSC line (19-9-11 iPSC). We then designed four experimental groups of co-cultured iPSC-CMs and iPSC-CFs: non-diseased control co-culture, LMNA+/E203K co-culture, non-diseased control CM/ LMNA+/E203K CF, and LMNA+/E203K CM/healthy CF.

Mechanical function was quantified with digital image correlation (DIC) to extract contractile kinetics while electrophysiology was assessed using optical mapping (OM). Culturing mutant fibroblasts with healthy myocytes accelerated the spontaneous beating rate, shortened time to peak contraction, increased calcium transient amplitude, and accelerated calcium handling dynamics. Pairing mutant myocytes with healthy fibroblasts mitigated the slow contractile kinetics and spontaneous beating rates observed in full-mutant co-cultures, while partially recovering calcium rise and decay times.

These results demonstrate that iPSC- CM and CF crosstalk modulates the functional phenotype of LMNA+/E203K mutations. Future transcriptomic analysis will provide insights to driving mechanisms behind calcium handling and contractile behavior.

Cardiovascular disease is the global leading cause of death and highlights the need for improved human relevant in vitro systems. Human pluripotent stem cells (hPSCs) provide a human relevant platform to study cardiac development and disease. Cardiac fibroblasts are crucial for heart tissue repair through generating extracellular matrix (ECM) proteins. Prior research demonstrates that human cardiac fibroblast subtypes (primary and hSPC) have functional differences, and disease models can be improved by understanding these differences. Cardiac fibroblasts are activated by chemical triggers such as TGF-β to repair tissue damage; however, overactivation can lead to fibrosis, which impairs cardiac function. The activation pathway upregulates glycolysis, causing distinct metabolic changes.

These metabolic differences can be characterized through two-photon fluorescence lifetime imaging microscopy (FLIM) of NAD(P)H and FAD, key metabolic coenzymes. Only the reduced NAD(P)H and oxidized FAD are autofluorescent and exhibit different fluorescence lifetimes based on protein binding state. FLIM was used to capture metabolic differences between ventricular, atrial, and hPSC cardiac fibroblasts. To further understand metabolic differences between subtypes, fibrotic markers were used to identify activation during stimulation. Fibroblasts were stimulated with TGF-β and treated with the TGF-β receptor inhibitor SB-431542 to suppress the activation pathway. Epifluorescence microscopy with immunofluorescent antibody staining captured changes in 𝛼SMA expression, which indicates activated fibroblasts.

The results showed atrial and ventricular fibroblasts had comparable metabolic states with similar NAD(P)H mean lifetimes. HPSC-derived cardiac fibroblasts had a higher NAD(P)H mean lifetime than primary fibroblasts, suggesting a lower glycolytic baseline. Following stimulation, the 𝛼SMA expression trend varied between cardiac subtypes. For the SB inhibitor + TGF-β condition, 𝛼SMA expression increased in the hPSC subtype compared to atrial and ventricular, indicating different functionality.

INTRODUCTION: Acute cellular rejection (ACR) is a common complication after lung transplantation that results in immune-mediated injury to the lung allograft and worsens patient outcomes. We sought to identify mechanisms of epithelial injury downstream of immune activation that may inform future therapeutic strategies. Focusing on the airway epithelium as an epicenter of injury in rejection, we generated airway organoids from the bronchoalveolar lavage (BAL) fluid of patients with and without ACR.

METHODS: BAL was collected from 5 lung transplant patients with biopsy-proven ACR and 3 patients without rejection. Using established methods, we generated airway organoids and expanded them for 2-5 passages. BAL-derived organoids were analyzed by scRNA-seq using standard Seurat packages. Differentially expressed genes were validated using RT-qPCR in an independent cohort of 8 organoid samples (5 ACR, 3 control).

RESULTS: Airway organoids derived from BAL samples contained four major epithelial populations: basal, cycling basal, suprabasal, and secretory cells. Comparative single-cell transcriptomic analysis demonstrated broad upregulation of inflammatory pathways and downregulation of homeostatic epithelial processes in organoids from patients with ACR. Genes shared across two or more epithelial clusters were prioritized for validation. RT-qPCR confirmed upregulation of genes involved in immunoregulatory signaling (HLA-DRB1, MUC1, ARTN), epithelial repair (KRT5, KRT8, POU6F2), and fibrosis (MUC4, NETO1).

DISCUSSION: Our results share similarities to recent transcriptomics from patient samples with various forms of rejection, indicating that our organoids can accurately reflect disease phenotypes. Future studies include further pathway investigation using our airway organoid models to identify potential therapeutic targets.

To address this limitation, we developed an automated image-analysis pipeline using custom-trained Cellpose-SAM models to detect, count, and measure lung organoids. Cellpose-SAM is a generalist algorithm commonly used for segmenting, or identifying boundaries between, cells. However, the default model was unable to segment organoids accurately. Because organoid morphology changes across passages, we trained two separate models. The early-passage model was optimized for images containing small developing structures, where distinguishing newly formed organoids from individual cells and cellular debris is challenging. A second, more generalized later-passage model was trained to identify well-established organoids of varying sizes. Following segmentation, individual organoids are automatically counted and measured. These outputs allow assessment of organoid growth across passages and experimental conditions.

We have applied the models to patient-derived lung organoid cultures to study acute cellular rejection (ACR) after lung transplantation. Quantifying both OFE and organoid size measurements allows us to capture complementary aspects of epithelial progenitor behavior, and may provide a more detailed picture of the epithelial injury and repair response associated with ACR. By reducing the image-analysis bottleneck, automated quantification of organoids will facilitate higher-throughput comparisons across patients, passages, and time points.

Methods: We expanded the organoids to passage 2, enzymatically digested the organoids to a single cell suspension, and performed short tandem repeat (STR) analysis. STR identifies the number of repeated sequences at loci across the genome, revealing an individual’s unique genetic signature. By comparing the organoids’ STR profile to a control containing only recipient-derived cells, we quantified the relative contributions of recipient- and donor-derived cells. We also performed STR analysis on BAL fluid, which predominantly contains immune rather than epithelial cells.

Results: We analyzed 7 lines of alveolar organoids and found that they were donor-derived. We analyzed 12 lines of airway organoids and found that some samples were donor-derived, some recipient-derived, and some a mixture. We also analyzed 9 lines of BAL immune cells which contained predominantly recipient-derived cells.

Discussion: Our results supported the hypothesis that the epithelial organoids taken from inside the transplanted lung contain mostly donor-derived cells while the BAL immune cells contain mostly recipient-derived cells. However, there was patient-to-patient variation. Thus, we can obtain epithelial and immune cells from two individuals from a single transplant BAL sample. Future work will include developing a method to isolate and culture pure populations of donor and recipient cells, followed by establishing a system to co-culture epithelial and immune cells to model lung transplant rejection.

Carmella Whittaker, Tatum Hedrick, Frank Seipel, Randolph S. Ashton

Segmenting and quantifying microscopy images of immunostained 3D tissues often requires protracted manual labor that is prone to quantitative inconsistency and biases. With AI-based segmentation models such as Cellpose-SAM, 3D image segmentation and quantification can be scaled while maintaining analytical consistency. This poster describes progress toward developing an algorithm that manages image processing, Cellpose-SAM segmentation, and quantification of nuclear and cytoplasmic cells features in immunostained human pluripotent stem cell (hPSC)-derived neural organoids. Validation of nuclear and cytoplasm stain masking models, integration with high-throughput computing clusters, and development of algorithmic spatial analysis will facilitate neural organoid research by making quantitative cytoarchitectural analysis more scalable, precise, and reproducible.

Carmella Whittaker acknowledges the support of NIH/NIGMS 5R01GM154933, “Guided Intracellular Delivery using Precise Area Introduction and Transfection (PAINT)”, and NSF Engineering Research Center for Cell Manufacturing Technologies (CMaT) #1648035 in producing the data analyzed on this poster.

To bridge the gap between computational candidate design and scalable experimental validation, we established an integrated ASO development pipeline paired with the RosetteArray™ platform designed by the Ashton Lab. The RosetteArray™ utilizes human pluripotent stem cell (hPSC)-derived neural rosettes in up to a 96-well format, effectively modeling early central nervous system development to enable reproducible, high-throughput phenotypic screening of neurodevelopmental disorders (NDDs). By combining sequence-level optimization with human-relevant neural tissue models, this framework rapidly prioritizes high-potential ASO designs prior to downstream testing.

As proof of concept, the foundational workflow established by this pipeline has been leveraged to evaluate splice-switching ASO candidates targeting monogenic NDDs, restoring functional cellular phenotypes in neural organoid systems as seen with Timothy Syndrome type 1. Future work will deploy the RosetteArray™ platform to run automated, high-throughput screens, validate ASO efficacy across diverse NDD targets, and continuously refine the therapeutic design workflow.

In addition, we are also exploring the immune-β-cell interface to better understand the root cause of T1D and identify key players involved in β-cell demise during an autoimmune attack. We hypothesize that T1D β-cells exhibit key differences in how they handle stress compared to non-diabetic β-cells.

Methods: DKO;iCP hESCs, HUES8 (WT) hESCs, and T1D iPSCs were differentiated into SCIs. SCIs were co-cultured with human spleen-derived, T cell-enriched activated splenocytes (EAS) or isolated natural killer (NK) cells. Granzyme B secretion was measured by ELISA after 24-72 hours to assess immune-mediated cytotoxicity. Co-cultures were repeated using ER-stressed SCIs induced with 250-500 nM thapsigargin for 24-48 hours, depending on cell type. ER stress was confirmed by flow cytometry.

Results: DKO;iCP, HUES8, and T1D SCIs displayed morphology comparable to cadaveric islets. DKO;iCP SCIs exhibited reduced autoreactive T-cell-mediated cytotoxicity compared with WT SCIs and reduced NK cell-mediated cytotoxicity compared with cadaveric islets. T1D-derived SCIs showed significantly greater T-cell-mediated cytotoxicity than ER-stressed WT SCIs, suggesting intrinsic disease-associated differences in β-cell state influence immune recognition and susceptibility to immune attack.

Discussion: Gene-edited SCIs reduce innate and adaptive immune-mediated cytotoxicity in vitro, while T1D-derived SCIs display enhanced immune susceptibility independent of engineered immune-evasive strategies. These findings improve our understanding of β-cell–immune interactions and support development of next-generation cell replacement therapies for T1D.

To elucidate the role of inhibitory GPCR signaling in islet dysfunction and diabetes disease progression, we knocked out Gnaz in an obesity-induced diabetes mouse model (BTBR OB mice). We found that Gnaz-null OB mice are protected from hyperglycemia and have an increased capacity for insulin secretion. IF imaging of pancreas sections shows juvenile Gnaz-null mice OB mice have higher proliferation of beta cells compared to WT OB siblings. As a result, adult Gnaz-null OB mice have doubled percent beta cell area compared to WT OB and all lean littermates. Loss of Gnaz increases GLP1 synthesis and secretion. We observed that over 50% WT OB islets have abolished islet architecture and acinar infiltration–commonly observed in pre-diabetic and diabetic patients. Gnaz-null OB islets maintain regular architecture and exhibit minimal acinar infiltration. Ucn3 staining to evaluate whether abolished islet architecture is due to loss of beta cell maturity revealed that Ucn3 is co-expressed in only 30% of insulin-positive beta cells in WT OB mice and 50% of insulin-positive beta cells in Gnaz-null OB mice, indicating a partial rescue.

Our findings highlight the key role that inhibitory G-proteins play in diabetes progression and reveal a therapeutic target to improve both endogenous GLP1 production and response to GLP1R signaling and increase the capacity for beta cell expansion and islet function.

Authors: Srishti Goswami1,2,3,4, Aindrila Saha2,3,4, Agustin Luz Madrigal1,3, Celia Bisbach1,3, Praveen Susaimanickam1,3, Yashdeep Phanse1,3, David Gamm1,3, Mrinalini Hoon1,2,3,4, Raunak Sinha1,2,3,4

1: Department of Ophthalmology and Visual Sciences, UW-Madison
2: Department of Neuroscience, UW-Madison
3: McPherson Eye Research Institute, UW-Madison
4: Cellular and Molecular Biology Graduate Program, UW-Madison

Human pluripotent stem cell (hPSC)-derived retinal organoids (ROs) represent a transformative advancement, offering a 3D system to recapitulate the complexity of the developing human retina. They serve as robust platforms for modelling inherited retinal diseases and testing therapies. The ability of RO photoreceptors to successfully relay sensory information depends on the formation of functional synapses to facilitate downstream signalling between photoreceptors and second-order neurons. Thus, validating the expression and function of photoreceptor synaptic proteins is critical. One family of synaptic proteins that facilitates neurotransmitter release from photoreceptor terminals are calcium channels but their physiological properties and expression profiles of these channels within RO photoreceptors remains unknown. This information is required as a key step for establishing ROs as authentic models of human retina.
We employed patch-clamp electrophysiology to measure calcium currents from RO photoreceptors and characterized the functional state of the organoid cone photoreceptors. Cone photoreceptors in primate retina support bright light and colour vision; mutations in their calcium channels lead to vision disorders. We benchmarked our findings of RO cone photoreceptor calcium channel properties against a gold-standard reference of fetal and adult macaque cone photoreceptors.
Our findings indicate that a subset of organoid cone photoreceptors exhibit synaptic function similar to fetal macaque. We are also planning to compare the trajectory of calcium channel functional profiles across RO photoreceptor developmental timepoints, correlating these with their light response profile. By aligning physiological profiles of RO photoreceptors across development with primate retinal physiology, this work establishes a rigorous framework for assessing the authenticity of lab-generated ROs. These findings are significant for generating disease-relevant models for therapeutic screening and drug discovery.

Specialized and functionally diverse epithelial cells mediate essential roles of epithelial tissues, including chemical sensing, immune cell interactions, and tissue homeostasis. However, the diverse specialized epithelial populations and their molecular signatures in the zebrafish fin epithelium remain poorly understood. Here, we integrated single-cell RNA sequencing (scRNA-seq) datasets encompassing uninjured and regenerating fins and identified specialized epithelial populations with distinct marker genes. Among them, we discovered a previously unrecognized tuft-like epithelial cell type marked by advillin (avil) expression. Using an avil reporter line, we uncovered the unique morphology and position of these cells: large epithelial cells projecting an apical, actin-enriched spike outward from the body – which we term spike cells. Notably, these avil+ spike cells exhibit close spatial association with a subset of mpeg1+ resident macrophages. During fin regeneration, these avil+ spike cells undergo dynamic changes in density, morphology, and orientation. Our data further suggest that these avil⁺ spike cells are primarily replenished through stem cell–mediated regeneration rather than self-renewal. As these cells mature, they progressively develop apical protrusions extending beyond the epithelial surface and expand their size from deeper epithelial layers toward the superficial epithelial layer, indicating a coordinated maturation process. Transcriptomic and pharmacological analyses revealed that spike cell regeneration depends on Alk signaling. Moreover, selective ablation of avil+ cells resulted in a marked reduction in macrophage abundance, suggesting a role of spike cells in maintaining the local immune cell population. Together, these findings establish a foundational framework for investigating tuft-like epithelial cell biology epithelial–immune interactions in zebrafish.

During mouse embryonic development ~40 cells are designated as Primordial Germ Cells (PGCs) at embryonic day 7.5 (E7.5) and are destined to become sperm or egg. These cells undergo massive epigenetic reprogramming of repressive histone and DNA methylation, which is not seen in surrounding somatic cells. The repressive histone modification H3K9me2 is depleted immediately after PGC specification and lost globally by E9.5. H3K9me2 can be added by methyltransferases, G9a and GLP, and removed by demethylases, KDM3A and KDM3B. The global depletion of H3K9me2 has been attributed to the downregulation of the G9a and GLP. However, Kdm3a and Kdm3b remain highly expressed in the PGCs at E11.5 – long after their catalytic target is gone. Both KDM3A and KDM3B are vital for mouse spermatogenesis and KDM3B is also important for female fertility. Yet, their role in germ cell development before sex specification remains unknown. Therefore, I employ an in vitro differentiation model of embryonic stem cells to primordial germ cell like cells (PGCLCs) in which we can rapidly eliminate the KDM3 proteins using a “degron” tag. The PGCLC differentiation is a directed embryoid body differentiation in which ~20% of the cells are PGCLCs. Since PGCLC differentiation is heterogenous, we used single cell Multiome sequencing to tease out the gene expression and chromatin profiles between wildtype and degron PGCLCs and the surrounding niche of cells. Using these tools, we found that retention of H3K9me2, through complete removal of the KDM3 proteins, compromises PGCLC differentiation.